A few things about the voting story I wrote that published today at Wired News.
First I want to clarify the quote from Kathy Dent, which I didn't have space to do in the story (I was working against a word limit). Florida Election Supervisor Kathy Dent told me that she was unaware of voter problems with the touch-screen machines until after the first tabulations were made and it was evident that some 18,000 ballots didn't have any votes cast in the congressional race. When I mentioned that press reports before the election indicated that voters had been having problems getting machines to register their selections during pre-voting, she said she learned about those only after someone from the Herald Tribune called her. When Dent told the Trib that no one had contacted her office about the problems, the Trib person, according to Dent, told the voters to contact Dent, which they did.
Dent called the voters who complained "voting activists" as in "yes there were a couple of people who complained to the Herald Tribune but they were voting activists." When I asked if she considered the complaints unreliable because of who made them she said, "I'm dismissing their particular reports. We checked the machines, and I checked with all my early-voting sites, and I didn’t get any indication that we were having a problem." So she had heard about some problems with the machines but was satisfied after checking with people running the early voting that the machines were fine. It was only after the election, she says, that other complaints came in.
Here's a Herald Trib story about the early voting complaints.
There's one other thing I want to point out about the FSU report mentioned in the story. The report states that the testers examined the source code and didn't do hands-on testing of the machines themselves. However, on pages 20-21 it mentions that other testers did examine the machines -- once looking at machines that weren't used in the November election and once doing test voting on machines that were used in the election and were responsible for some undervotes. Those tests were conducted by a different group organized by the state. You'll find their report here. Page 2 and Appendix C discuss that hands-on testing.
One thing I should note: The testers refer to it as "parallel testing" but this is a misnomer since parallel testing in the elections realm is generally understood to describe testing that occurs on Election Day with machines pulled offline at the last minute. Election officials will pull aside a certain number of machines the morning of the election (or a day before -- the optimal time depends on when machines are delivered to polling places since you want to make sure that the remaining machines aren't sitting around somewhere where they can still be rigged before the election). Examiners will test the sample machines to make sure they operate as they should. Testing the machines days after an election, as the Florida testers did, requires that officials keep those machines under lock and key with careful chain-of-custody tracking to ensure that no one tampers with them.
With regard to the merits of the reports, some qualified computer scientists are currently examining them and preparing a paper. No date yet on when their paper will be available.